<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=utf-8"><meta name=Generator content="Microsoft Word 15 (filtered medium)"><!--[if !mso]><style>v\:* {behavior:url(#default#VML);}
o\:* {behavior:url(#default#VML);}
w\:* {behavior:url(#default#VML);}
.shape {behavior:url(#default#VML);}
</style><![endif]--><style><!--
/* Font Definitions */
@font-face
{font-family:Helvetica;
panose-1:2 11 5 4 2 2 2 2 2 4;}
@font-face
{font-family:SimSun;
panose-1:2 1 6 0 3 1 1 1 1 1;}
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:DengXian;
panose-1:2 1 6 0 3 1 1 1 1 1;}
@font-face
{font-family:DengXian;
panose-1:2 1 6 0 3 1 1 1 1 1;}
@font-face
{font-family:SimSun;
panose-1:2 1 6 0 3 1 1 1 1 1;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0cm;
text-align:justify;
text-justify:inter-ideograph;
font-size:10.5pt;
font-family:DengXian;}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
p.MsoListParagraph, li.MsoListParagraph, div.MsoListParagraph
{mso-style-priority:34;
margin:0cm;
text-align:justify;
text-justify:inter-ideograph;
text-indent:21.0pt;
font-size:10.5pt;
font-family:DengXian;}
.MsoChpDefault
{mso-style-type:export-only;}
/* Page Definitions */
@page WordSection1
{size:612.0pt 792.0pt;
margin:72.0pt 90.0pt 72.0pt 90.0pt;}
div.WordSection1
{page:WordSection1;}
/* List Definitions */
@list l0
{mso-list-id:874082994;
mso-list-type:hybrid;
mso-list-template-ids:-436729666 -1 67698713 67698715 67698703 67698713 67698715 67698703 67698713 67698715;}
@list l0:level1
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:18.0pt;
text-indent:-18.0pt;}
@list l0:level2
{mso-level-number-format:alpha-lower;
mso-level-text:"%2\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:42.0pt;
text-indent:-21.0pt;}
@list l0:level3
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:63.0pt;
text-indent:-21.0pt;}
@list l0:level4
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:84.0pt;
text-indent:-21.0pt;}
@list l0:level5
{mso-level-number-format:alpha-lower;
mso-level-text:"%5\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:105.0pt;
text-indent:-21.0pt;}
@list l0:level6
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:126.0pt;
text-indent:-21.0pt;}
@list l0:level7
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:147.0pt;
text-indent:-21.0pt;}
@list l0:level8
{mso-level-number-format:alpha-lower;
mso-level-text:"%8\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:168.0pt;
text-indent:-21.0pt;}
@list l0:level9
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:189.0pt;
text-indent:-21.0pt;}
@list l1
{mso-list-id:1015612098;
mso-list-type:hybrid;
mso-list-template-ids:467171602 -1 67698713 67698715 67698703 67698713 67698715 67698703 67698713 67698715;}
@list l1:level1
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:18.0pt;
text-indent:-18.0pt;}
@list l1:level2
{mso-level-number-format:alpha-lower;
mso-level-text:"%2\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:42.0pt;
text-indent:-21.0pt;}
@list l1:level3
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:63.0pt;
text-indent:-21.0pt;}
@list l1:level4
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:84.0pt;
text-indent:-21.0pt;}
@list l1:level5
{mso-level-number-format:alpha-lower;
mso-level-text:"%5\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:105.0pt;
text-indent:-21.0pt;}
@list l1:level6
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:126.0pt;
text-indent:-21.0pt;}
@list l1:level7
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:147.0pt;
text-indent:-21.0pt;}
@list l1:level8
{mso-level-number-format:alpha-lower;
mso-level-text:"%8\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:168.0pt;
text-indent:-21.0pt;}
@list l1:level9
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:189.0pt;
text-indent:-21.0pt;}
@list l2
{mso-list-id:1431967431;
mso-list-type:hybrid;
mso-list-template-ids:644408836 -1 67698713 67698715 67698703 67698713 67698715 67698703 67698713 67698715;}
@list l2:level1
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:18.0pt;
text-indent:-18.0pt;}
@list l2:level2
{mso-level-number-format:alpha-lower;
mso-level-text:"%2\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:42.0pt;
text-indent:-21.0pt;}
@list l2:level3
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:63.0pt;
text-indent:-21.0pt;}
@list l2:level4
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:84.0pt;
text-indent:-21.0pt;}
@list l2:level5
{mso-level-number-format:alpha-lower;
mso-level-text:"%5\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:105.0pt;
text-indent:-21.0pt;}
@list l2:level6
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:126.0pt;
text-indent:-21.0pt;}
@list l2:level7
{mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:147.0pt;
text-indent:-21.0pt;}
@list l2:level8
{mso-level-number-format:alpha-lower;
mso-level-text:"%8\)";
mso-level-tab-stop:none;
mso-level-number-position:left;
margin-left:168.0pt;
text-indent:-21.0pt;}
@list l2:level9
{mso-level-number-format:roman-lower;
mso-level-tab-stop:none;
mso-level-number-position:right;
margin-left:189.0pt;
text-indent:-21.0pt;}
ol
{margin-bottom:0cm;}
ul
{margin-bottom:0cm;}
--></style></head><body lang=ZH-CN link=blue vlink="#954F72" style='word-wrap:break-word'><div class=WordSection1><div><div><p class=MsoNormal align=left style='text-align:left'><span lang=EN-US>Hi Adapters,</span><span lang=EN-US style='font-size:12.0pt'><o:p></o:p></span></p></div></div><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US>In the talk about chain of thought this afternoon, there are some details that are not clear. Here are the supplements.</span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoListParagraph style='margin-left:18.0pt;text-indent:-18.0pt;mso-list:l1 level1 lfo3'><![if !supportLists]><span lang=EN-US><span style='mso-list:Ignore'>1.<span style='font:7.0pt "Times New Roman"'> </span></span></span><![endif]><span lang=EN-US>For the promptings used in the accuracy test, it is narrated in the paper like this: “As most of the datasets only have an evaluation split, we manually composed a single set of eight generic few-shot exemplars with chains of thought for promptingFigure 1 (bottom) shows one chain of thought exemplar, and the full set of exemplars is given in <span style='background:yellow;mso-highlight:yellow'>Table 20</span> in the Appendix. (These exemplars did not undergo prompt engineering, and we assess robustness to alternative exemplars and chains of thought in <span style='color:#70AD47'>Section 3.4.</span>) To investigate whether chain of thought prompting in this form can successfully elicit successful reasoning across a wide range of math word problems, we used this single set of eight chain of thought exemplars for all six datasets.” <span style='background:yellow;mso-highlight:yellow'>Table 20 is shown below. </span>And <span style='color:#70AD47'>Section3.4</span> is the chapter about ablation test, which changes the annotator, the examplar set and the propmting style in the experiment, and this part has been introduced in the talk.</span></p><p class=MsoNormal><span lang=EN-US><img width=269 height=624 style='width:2.8in;height:6.5in' id="图片_x0020_3" src="cid:image003.png@01D8EF04.BA27AE60"></span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US>For the symbolic reasoning tasks, it is narrated like this: “We again manually compose chains of thought for the few-shot exemplars for each task, which are shown in <span style='background:yellow;mso-highlight:yellow'>Table 4</span> on the next page.” <span style='background:yellow;mso-highlight:yellow'>Table 4</span> is shown below. The full prompts with all examplars are also 8-pair sets, if you are interested in it, you can look for it in the paper.</span></p><p class=MsoNormal><span lang=EN-US><img width=383 height=421 style='width:3.9916in;height:4.3833in' id="图片_x0020_4" src="cid:image005.png@01D8EF04.BA27AE60"></span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US>For commonsence reasoning, it is narrated like this: “For CommonsenseQA and StrategyQA, we randomly selected eight examples from the training set and manually composed chains of thought for them to use as few-shot exemplars. The two BIG-bench tasks do not have training sets, so we selected the first ten examples as exemplars in the evaluation set as few-shot exemplars and report numbers on the rest of the evaluation set.”</span></p><p class=MsoNormal><span lang=EN-US>Here is a brief example: </span></p><p class=MsoNormal><span lang=EN-US><img width=383 height=568 style='width:3.9916in;height:5.9166in' id="图片_x0020_5" src="cid:image008.png@01D8EF04.BA27AE60"></span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoListParagraph style='margin-left:18.0pt;text-indent:-18.0pt;mso-list:l1 level1 lfo3'><![if !supportLists]><span lang=EN-US><span style='mso-list:Ignore'>2.<span style='font:7.0pt "Times New Roman"'> </span></span></span><![endif]><span lang=EN-US>About the publication of the papers. </span></p><p class=MsoListParagraph style='margin-left:42.0pt;text-indent:-21.0pt;mso-list:l1 level2 lfo3'><![if !supportLists]><span lang=EN-US><span style='mso-list:Ignore'>a)<span style='font:7.0pt "Times New Roman"'> </span></span></span><![endif]><span lang=EN-US>The main paper which introduces CoT: “Chain of Thought Prompting Elicits Reasoning in Large Language Models”. This paper has not been accepted yet, but it has been cited 99 times till now. This paper is currently updating, the newest version is V5(</span><span lang=EN-US style='font-size:9.5pt;font-family:"Helvetica",sans-serif;color:black;background:white'>Mon, 10 Oct 2022 20:21:17 UTC</span><span lang=EN-US>) which may contains more experiment. My talk is based on V2(</span><span lang=EN-US style='font-size:9.5pt;font-family:"Helvetica",sans-serif;color:black;background:white'>Wed, 6 Apr 2022 03:51:50 UTC</span><span lang=EN-US>).</span></p><p class=MsoListParagraph style='margin-left:42.0pt;text-indent:-21.0pt;mso-list:l1 level2 lfo3'><![if !supportLists]><span lang=EN-US><span style='mso-list:Ignore'>b)<span style='font:7.0pt "Times New Roman"'> </span></span></span><![endif]><span lang=EN-US>The paper which introduces zero-shot-CoT is ” Large Language Models are Zero-Shot Reasoners”. This paper has not been accepted also, with citation 44. It is also in an update period. My talk is based on the latest version </span><strong><span lang=EN-US style='font-size:9.5pt;font-family:"Helvetica",sans-serif;color:black;background:white'>[v3]</span></strong><span lang=EN-US style='font-size:9.5pt;font-family:"Helvetica",sans-serif;color:black;background:white'> Sun, 2 Oct 2022 07:12:50 UTC</span></p><p class=MsoListParagraph style='margin-left:42.0pt;text-indent:-21.0pt;mso-list:l1 level2 lfo3'><![if !supportLists]><span lang=EN-US><span style='mso-list:Ignore'>c)<span style='font:7.0pt "Times New Roman"'> </span></span></span><![endif]><span lang=EN-US style='color:black;background:white'>The paper which introduces auto-CoT is “Automatic Chain of Thought Prompting in Large Language Models”. This paper also has not been accepted yet. My talk is based on the latest version </span><strong><span lang=EN-US style='font-size:9.5pt;font-family:"Helvetica",sans-serif;color:black;background:white'>[v1]</span></strong><span lang=EN-US style='font-size:9.5pt;font-family:"Helvetica",sans-serif;color:black;background:white'> Fri, 7 Oct 2022 12:28:21 UTC</span><span lang=EN-US><o:p></o:p></span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US>Hope you enjot it!</span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US>Best regards, </span></p><p class=MsoNormal><span lang=EN-US>Chunhao</span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoNormal><span lang=EN-US><o:p> </o:p></span></p><p class=MsoNormal>从<span lang=EN-US> Windows </span>版<span lang=EN-US><a href="https://go.microsoft.com/fwlink/?LinkId=550986"><span lang=EN-US><span lang=EN-US>邮件</span></span></a></span>发送</p><p class=MsoNormal><span lang=EN-US style='font-size:12.0pt;font-family:SimSun'><o:p> </o:p></span></p></div></body></html>